skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


Title: Enhancing Urban Flow Maps via Neural ODEs
Flow super-resolution (FSR) enables inferring fine-grained urban flows with coarse-grained observations and plays an important role in traffic monitoring and prediction. The existing FSR solutions rely on deep CNN models (e.g., ResNet) for learning spatial correlation, incurring excessive memory cost and numerous parameter updates. We propose to tackle the urban flows inference using dynamic systems paradigm and present a new method FODE -- FSR with Ordinary Differential Equations (ODEs). FODE extends neural ODEs by introducing an affine coupling layer to overcome the problem of numerically unstable gradient computation, which allows more accurate and efficient spatial correlation estimation, without extra memory cost. In addition, FODE provides a flexible balance between flow inference accuracy and computational efficiency. A FODE-based augmented normalization mechanism is further introduced to constrain the flow distribution with the influence of external factors. Experimental evaluations on two real-world datasets demonstrate that FODE significantly outperforms several baseline approaches.  more » « less
Award ID(s):
1823279
PAR ID:
10211118
Author(s) / Creator(s):
; ; ; ; ;
Date Published:
Journal Name:
Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence, {IJCAI} 2020
Page Range / eLocation ID:
1295 to 1302
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; ; ; (, International Joint Conferences on Artificial Intelligence Organization)
    Fine-grained urban flow inference (FUFI), which involves inferring fine-grained flow maps from their coarse-grained counterparts, is of tremendous interest in the realm of sustainable urban traffic services. To address the FUFI, existing solutions mainly concentrate on investigating spatial dependencies, introducing external factors, reducing excessive memory costs, etc., -- while rarely considering the catastrophic forgetting (CF) problem. Motivated by recent operator learning, we present an Urban Neural Operator solution with Incremental learning (UNOI), primarily seeking to learn grained-invariant solutions for FUFI in addition to addressing CF. Specifically, we devise an urban neural operator (UNO) in UNOI that learns mappings between approximation spaces by treating the different-grained flows as continuous functions, allowing a more flexible capture of spatial correlations. Furthermore, the phenomenon of CF behind time-related flows could hinder the capture of flow dynamics. Thus, UNOI mitigates CF concerns as well as privacy issues by placing UNO blocks in two incremental settings, i.e., flow-related and task-related. Experimental results on large-scale real-world datasets demonstrate the superiority of our proposed solution against the baselines. 
    more » « less
  2. ; ; (, MobiCom '22: Proceedings of the 28th Annual International Conference on Mobile Computing And Networking)
    Convolutional neural networks (CNNs) play an important role in today's mobile and edge computing systems for vision-based tasks like object classification and detection. However, state-of-the-art methods on CNN acceleration are trapped in either limited practical latency speed-up on general computing platforms or latency speed-up with severe accuracy loss. In this paper, we propose a spatial-based dynamic CNN acceleration framework, NeuLens, for mobile and edge platforms. Specially, we design a novel dynamic inference mechanism, assemble region-aware convolution (ARAC) supernet, that peels off redundant operations inside CNN models as many as possible based on spatial redundancy and channel slicing. In ARAC supernet, the CNN inference flow is split into multiple independent micro-flows, and the computational cost of each can be autonomously adjusted based on its tiled-input content and application requirements. These micro-flows can be loaded into hardware like GPUs as single models. Consequently, its operation reduction can be well translated into latency speed-up and is compatible with hardware-level accelerations. Moreover, the inference accuracy can be well preserved by identifying critical regions on images and processing them in the original resolution with large micro-flow. Based on our evaluation, NeuLens outperforms baseline methods by up to 58% latency reduction with the same accuracy and by up to 67.9% accuracy improvement under the same latency/memory constraints. 
    more » « less
  3. ; ; ; ; (, Hydrology and Earth System Sciences)
    Abstract. The impact of spatial and temporal variability of rainfall on hydrological response remains poorly understood, in particular in urban catchments due to their strong variability in land use, a high degree of imperviousness and the presence of stormwater infrastructure. In this study, we analyze the effect of storm scale, position and movement in relation to basin scale and flow-path network structure on urban hydrological response. A catalog of 279 peak events was extracted from a high-quality observational dataset covering 15 years of flow observations and radar rainfall data for five (semi)urbanized basins ranging from 7.0 to 111.1 km2 in size. Results showed that the largest peak flows in the event catalog were associated with storm core scales exceeding basin scale, for all except the largest basin. Spatial scale of flood-producing storm events in the smaller basins fell into two groups: storms of large spatial scales exceeding basin size or small, concentrated events, with storm core much smaller than basin size. For the majority of events, spatial rainfall variability was strongly smoothed by the flow-path network, increasingly so for larger basin size. Correlation analysis showed that position of the storm in relation to the flow-path network was significantly correlated with peak flow in the smallest and in the two more urbanized basins. Analysis of storm movement relative to the flow-path network showed that direction of storm movement, upstream or downstream relative to the flow-path network, had little influence on hydrological response. Slow-moving storms tend to be associated with higher peak flows and longer lag times. Unexpectedly, position of the storm relative to impervious cover within the basins had little effect on flow peaks. These findings show the importance of observation-based analysis in validating and improving our understanding of interactions between the spatial distribution of rainfall and catchment variability. 
    more » « less
  4. ; (, Measurement Science and Technology)
    Abstract Optical flow methods have been developed over the past two decades for application to PIV images with the goal of acquiring higher resolution measurements of the velocity field than conventional cross-correlation-based techniques. Numerous optical flow velocimetry (OFV) algorithms have been devised to solve the ill-posed optical flow problem, with various physics-inspired strategies to tailor them to fluid flows. While OFV can be applied to continuous scalar fields, it has demonstrated the most success on images of tracer particles, i.e.\ traditional planar PIV images. Compared to state-of-the-art cross-correlation algorithms, OFV methods have demonstrated an order of magnitude increase in spatial resolution and up to a factor of two improvement in overall accuracy when evaluated on synthetic data, at the cost of increased computational time. The requirements for particle seeding density, inter-frame displacement, and image quality are also more stringent for OFV methods compared to cross-correlation. OFV has been applied sparingly in experiments to date, but appears to offer the same advantages demonstrated on synthetic data. At this stage, OFV seems best suited to planar velocity measurements, although extensions to stereoscopic measurements have been demonstrated. 
    more » « less
  5. ; ; (, Environmental Research Letters)
    Abstract Food supply chains are essential for distributing goods from production to consumption points. These complex supply chains are important for food security and availability. Recent research has developed novel methods to estimate food flows with high spatial resolution, but we do not currently understand how fine-grained food supply chains vary in time. In this study, we use an improved version of the Food Flow Model to estimate food flows (kg) between all county pairs across all food commodity groups for the years 2007, 2012, and 2017 (which requires estimating 206.3 million links). We then determine the core counties to the US food flow networks through time with a multi-criteria decision analysis technique. Our estimates of county-to-county food flows in time are freely available with this paper and could be useful for future research, policy, and decision-making. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email